%0 Conference Paper %1 10.1007/978-3-319-71246-8_29 %A Cerqueira, Vítor %A Torgo, Luís %A Pinto, Fábio %A Soares, Carlos %B Machine Learning and Knowledge Discovery in Databases %C Cham %D 2017 %E Ceci, Michelangelo %E Hollmén, Jaakko %E Todorovski, Ljupco %E Vens, Celine %E Dzeroski, Saso %I Springer International Publishing %K calibration timeseries %P 478--494 %T Arbitrated Ensemble for Time Series Forecasting %X This paper proposes an ensemble method for time series forecasting tasks. Combining different forecasting models is a common approach to tackle these problems. State-of-the-art methods track the loss of the available models and adapt their weights accordingly. Metalearning strategies such as stacking are also used in these tasks. We propose a metalearning approach for adaptively combining forecasting models that specializes them across the time series. Our assumption is that different forecasting models have different areas of expertise and a varying relative performance. Moreover, many time series show recurring structures due to factors such as seasonality. Therefore, the ability of a method to deal with changes in relative performance of models as well as recurrent changes in the data distribution can be very useful in dynamic environments. Our approach is based on an ensemble of heterogeneous forecasters, arbitrated by a metalearning model. This strategy is designed to cope with the different dynamics of time series and quickly adapt the ensemble to regime changes. We validate our proposal using time series from several real world domains. Empirical results show the competitiveness of the method in comparison to state-of-the-art approaches for combining forecasters. %@ 978-3-319-71246-8

@inproceedings{10.1007/978-3-319-71246-8_29, abstract = {This paper proposes an ensemble method for time series forecasting tasks. Combining different forecasting models is a common approach to tackle these problems. State-of-the-art methods track the loss of the available models and adapt their weights accordingly. Metalearning strategies such as stacking are also used in these tasks. We propose a metalearning approach for adaptively combining forecasting models that specializes them across the time series. Our assumption is that different forecasting models have different areas of expertise and a varying relative performance. Moreover, many time series show recurring structures due to factors such as seasonality. Therefore, the ability of a method to deal with changes in relative performance of models as well as recurrent changes in the data distribution can be very useful in dynamic environments. Our approach is based on an ensemble of heterogeneous forecasters, arbitrated by a metalearning model. This strategy is designed to cope with the different dynamics of time series and quickly adapt the ensemble to regime changes. We validate our proposal using time series from several real world domains. Empirical results show the competitiveness of the method in comparison to state-of-the-art approaches for combining forecasters.}, added-at = {2019-04-09T10:29:40.000+0200}, address = {Cham}, author = {Cerqueira, V{\'i}tor and Torgo, Lu{\'i}s and Pinto, F{\'a}bio and Soares, Carlos}, biburl = {https://www.bibsonomy.org/bibtex/24513d822d4176e73aeb1cf7a205ea6f5/lautenschlager}, booktitle = {Machine Learning and Knowledge Discovery in Databases}, description = {Arbitrated Ensemble for Time Series Forecasting | SpringerLink}, editor = {Ceci, Michelangelo and Hollm{\'e}n, Jaakko and Todorovski, Ljup{\v{c}}o and Vens, Celine and D{\v{z}}eroski, Sa{\v{s}}o}, interhash = {d3b89b4039efa8d52c376e1a3e09a20e}, intrahash = {4513d822d4176e73aeb1cf7a205ea6f5}, isbn = {978-3-319-71246-8}, keywords = {calibration timeseries}, pages = {478--494}, publisher = {Springer International Publishing}, timestamp = {2019-04-09T10:29:40.000+0200}, title = {Arbitrated Ensemble for Time Series Forecasting}, year = 2017 }